Photo Credit: Kitsawet Saethao

The following is a summary of “Machine learning derived serum creatinine trajectories in acute kidney injury in critically ill patients with sepsis,” published in the May 2024 issue of Critical Care by Takkavatakarn et al.

Researchers conducted a retrospective study assessing whether classifying acute kidney injury (AKI) in patients with critically septic illness based on early patterns of creatinine changes rather than just peak levels could improve the prediction of outcomes.

They identified patients meeting Sepsis-3 criteria who developed AKI within 48 hours of ICU admission using the Medical Information Mart for Intensive Care-IV database. Employing latent class mixed modeling, early creatinine trajectory-based AKI classes were delineated in patients with critically septic illness. The primary focus was on developing acute kidney disease (AKD). Secondary endpoints included a composite of AKD or all-cause in-hospital mortality by day 7 and AKD or all-cause in-hospital mortality by discharge. Multivariable regression was utilized to evaluate the impact of creatinine trajectory-based classification on outcomes, with external validation conducted using the eICU database.

The results showed 4,197 AKI in patients with critically septic illness, and 8 creatinine trajectory-based classes were identified. Compared to transient AKI, the class with severe AKI but mild improvement yet persistence had the highest adjusted risks for developing AKD (OR 5.16; 95% CI 2.87–9.24) and composite 7-day outcome (HR 4.51; 95% CI 2.69–7.56). Late mild AKI with persistence and worsening had the highest risks for the composite hospital discharge outcome (HR 2.04; 95% CI 1.41–2.94). External validation of associations remained consistent.

Investigators concluded that classifying AKI in patients with critically septic illness based on eight distinct early creatinine trajectory patterns improved outcome prediction beyond traditional AKI severity staging.

Source: ccforum.biomedcentral.com/articles/10.1186/s13054-024-04935-x